DE69215369T2 - POSITION MEASUREMENT - Google Patents

Abstract

Apparatus for measuring surfaces or profiles or textures comprises a grating interferometer comprising a curved grating (300) carried on a pivoted support arm (120,123) carrying at its other end a probe (130) for contact with a surface, biased into contact therewith by an electro-magnetic coil (410) acting on an armature (420), or a pair of such biasing means. A laser diode (310) illuminates the grating (300) to produce a pair of first order diffracted beams of opposite sign which are reflected from internal faces of a prism (316), and combined by a dichroic central layer thereof (335) and a pair of beam splitters (340a,340b). The output signals from the beam splitters are supplied to a signal processing circuit comprising a fringe counter (600) and an interpolator (700), the fringe counter detecting zero crossings of the signals and the interpolator (700) maintaining a digital estimate of the phase of the signals and updating the estimate when the phase difference between the estimate and the input signals exceeds a predetermined threshold. The interpolator comprises a digital counter (721) the output of which comprises the low order bits of the digital output signal for which the output of the fringe counter comprises the high order bits. <IMAGE>

Description

The invention relates to a device for measuring surfaces and profiles using interferometry. The invention also relates to a method and a device for determining, for example, the deflection of a probe over such a surface or profile from the output signal of an interferometer.

It is known to use devices for measuring the surface or profile of an object in order to deduce the texture or roughness or shape thereof. An example of such devices is the FORMA TALYSURF (TM) measuring system available from Rank Taylor Hobson Limited, PO Box 36, 2 New Star Road, Leicester LE4 7JQ, UK. This device comprises a probe element, or pickup needle, projecting downwardly to contact the surface to be measured and located adjacent one end of a support arm which is pivotally attached to a support structure.

The support structure is mounted for linear movement parallel (or approximately parallel) to the surface of the object to be measured and is moved in a line across the surface by a drive system.

The support arm extends beyond the pivot mount and carries a reflective surface that forms one end of two optical paths along which a collimated light source (a laser) is directed.

At a reference probe position above the surface, the two path lengths are equal. However, as the surface height varies, gravity forces the probe to follow the surface and the reflecting surface at the other end of the rotating support arm moves, changing the optical path length of one of the paths and thus creating an interference fringe pattern.

The number of stripes passing through a given optical detector position provides a measure of the deflection of the probe. Therefore, a stripe counting device is provided to continuously count the stripes passing through a fixed detector position and to calculate a To generate an output signal that represents the probe deflection over the surface to be measured while the probe structure is moved linearly over the surface. Of course, it would also be possible to move the surface instead of the probe.

To provide a measurement of the roundness or profile of a rotating object such as a crankshaft or axle, the object is positioned under the pick-up needle and supported so that it can be driven in rotation. As the object rotates, the probe measures its circumference, from which any eccentricity or deviation from a desired profile can be detected.

Such a measuring device must have extremely high accuracy in measuring the probe position (and hence the height of the surface). The device described above can achieve a displacement resolution of the order of 10 mm. Another important property of such a device is the maximum displacement that the pickup needle can measure; this must be reasonably large to be able to measure many types of surfaces or profiles and is typically of the order of millimetres. A useful measure of the performance of such measuring equipment is the "dynamic range", which is defined as range R(mm)/resolution R(mm). This value should preferably be as high as possible.

While the above measuring device provides excellent performance, a number of problems can arise. First, the Michelson-type interferometer used measures the difference in optical path lengths between the two existing optical paths. This critically depends on a stable wavelength of light, and changes in atmospheric pressure and temperature can cause changes in the wavelength of light and thus lead to incorrect measurement results. Since the two optical paths can have very different lengths, the light source must have a very large coherence length; providing a suitable light source thus requires an expensive and bulky type of laser, which requires a high voltage supply and is accompanied by considerable heat radiation.

The document US 3,726,595, Fig. 8-1, shows a device for surface measurement that uses a grating interferometer instead. In a grating interferometer, a beam of light illuminates a grating and is diffracted to produce a pair of first-order diffracted beams (although higher orders could be used). The two beams are reflected so that they travel equal path lengths and they are recombined to produce an interference pattern. As the grating moves transversely, the path of each beam remains constant, but the phase of each beam changes so that the fringes of the interference pattern shift. The movement of the fringes therefore provides a measure of the transverse motion of the grating.

According to document US 3,726,595, the grating is positioned normal to a surface and carries a probe at its surface-contacting end, so that when the interferometer is moved across the surface, the grating is forced to move normal to the surface, the resulting change in the interference pattern providing a measure of the probe position.

The nature of this device requires that the grating should be restricted to move only linearly, transverse to the illumination beam and to the line separating the two diffracted beams.

However, when measuring rough or irregular surfaces, or generally surfaces where there are rising edges that the probe encounters, this method of probe attachment would be unsatisfactory because if the probe is drawn into contact with such a rising edge, compressive stresses would arise within the probe as the probe was moved over the edge, which would firstly tend to shift the grating alignment (disturb the interference pattern) and secondly tend to cause probe vibration due to increased friction with the surface. There would also be a tendency to impose increased stresses on the probe attachment.

If the device shown in US 3,726,595 were designed so that the grating/probe were on a rotating arm, the inclination of the grating would rotate as the probe, and hence the grating, rose and fell above the surface. This rotation would either prevent diffraction completely or alter the direction of the diffracted rays, either of which would render the device ineffective.

Document EP-A-0 242 436 describes a device for measuring small paths with a measuring tip attached to one end of a rotating arm. The other end of the arm carries a measuring mirror. The movement of the rotating arm is detected using an optoelectronic component. The optoelectronic component has a grating for coupling light out of a waveguide and back into it. The diffraction grating directs light from the waveguide down to the measuring mirror and couples light reflected from the measuring mirror back into the waveguide. The reflected light or measuring beam is coupled into the waveguide in a coupling region to interfere with a reference beam to produce an interference pattern that depends on the movement of the measuring tip.

Document US-A-3,419,330 describes a device for measuring the angular rotational speed of a shaft, wherein a diffraction grating is arranged on the cylindrical surface of a disk which is mounted so as to rotate with the shaft.

Document FR-A-2 075 408 describes a measurement method in which a diffraction grating is attached to an object to be measured and diffraction beams of different orders travel along optical paths of substantially the same length and through a beam combiner in which the polarized components are phase-shifted relative to each other. On leaving the beam combiner, the components are shifted relative to each other in terms of the modulation phase and detected by a photoelectric device to provide signals from which the displacement of the object can be deduced.

According to one aspect of the invention there is provided an apparatus for measuring surface features, comprising a probe for contacting a surface to be measured and moving along the surface, and a probe holder for supporting the probe for rotational movement about a pivot point, characterized by a grating interferometer for providing a measure of the rotational movement of the probe, a diffraction grating disposed on a curved surface connected to the probe holder for movement with the probe and having a center of curvature at the point about which the probe can rotate.

A device embodying the invention is for measuring rough surfaces with rising edges, it allows the optical path length of the diffracted rays to be kept constant and it allows the diffraction pattern to depend to a much smaller extent on the wavelength of the illuminating light and its coherence length, so that relatively cheap low-power light sources such as semiconductor lasers can be used.

The grid can be mounted with the support arm at the other end relative to the pivot connection with the probe. This extends the effective length of the support arm on the probe side of the pivot point, which can be advantageous when measuring enclosed volumes such as tubes.

However, the curvature of the grating in this embodiment causes the diffracted beams to diverge or converge depending on whether the grating is convex or concave, respectively. Therefore, means for optically correcting the divergence or convergence introduced by the grating is preferably provided. The means preferably comprises a correcting lens or it comprises correcting lenses. When a source of diverging light such as a semiconductor laser is used, the correcting means preferably also corrects the divergence of the light source. Alternatively, the divergence of the light source can be used to compensate for convergence due to the grating.

In the grating interferometer shown in Fig. 8-1 of document US-3,726,595, the two diffracted beams are arranged to travel along two paths of identical length by the provision of a pair of parallel, plane mirrors which redirect each beam to a beam splitter which combines the two reflected beams.

In one embodiment of the invention, an interferometric measuring device includes a prism arranged relative to a diffraction grating such that a pair of the sides of the prism receive, at their inner surfaces, a corresponding pair of rays diffracted by the grating of the same but opposite order and direct them along paths of preferably equal length to an optical compositing element for producing an interference pattern between them. Preferably, the optical compositing means (e.g. a beam splitter) also comprises part of the prism; preferably a beam splitter is formed as the central inner surface thereof. Preferably, the shape, position and Material of the prism is selected so that reflection occurs according to total internal reflection. By providing the optical components as part of a single prism, the number of alignment and calibration operations requiring high accuracy is greatly reduced, which reduces the production and maintenance costs of the device.

Preferably, the prism is designed so that the angle of incidence on these reflection surfaces is approximately 45°; this enables the use of a simpler assembly device.

The two reflection surfaces should be parallel to simplify the setup.

In the context of the TALYSURF (TM) interferometer discussed above, it was stated that the deflection was measured by counting the number of fringes passing through the position of an optical detector. This method provides an accurate measurement, but is limited by the step size of the fringes, which in turn is related to the grating step size. However, greater accuracy can be achieved by measuring the optical signal phase between fringes at the detector position; this is called interpolation between fringes.

Document US-4,629,886 describes an optical scale reader used in the position control of a VLSI manufacturing table carrying a scale or diffraction grating. The linear position of the table and scale is measured by a grating interferometer and there are two optical detectors providing signals 90º out of phase with each other. This device provides both fringe counting and an interpolator which operates by selecting which of the two signals is considered to be a better approximation of linearity (in the sense that a sinusoidal signal is an approximation of a linear signal for small values), digitizing this signal and using the digitized signal as a measure of the interpolation value.

However, the accuracy of this type of interpolator is limited unless a mathematical normalization step is performed for the two phase-shifted signals. Furthermore, it requires an analog-to-digital converter with reasonable accuracy. The analog-to-digital conversion and all subsequent Calculations limit the operating speed of the interpolator.

Accordingly, in one embodiment of the invention there is provided an interpolator circuit for use with an interferometric measuring device or the like, comprising means for generating a reference signal, means for generating a signal representing the difference between the phase of the reference signal and that of the signal derived from the interferometer, and means for varying the reference signal to reduce the difference. The reference signal thus follows the phase of the signal derived from the interferometer, and the phase of the reference signal is the output signal rather than the signal derived from the interferometer.

Preferably, the means for providing the reference signal comprises means for generating a control signal and means for providing, in response to the control signal, a signal of predetermined phase. This arrangement is preferred to an arrangement in which the control signal controls the frequency (or rate of phase change) because it is accurate even when the phase is constant (e.g. when the probe is stationary).

In the field of tracking rotating machines, it is known to use tracking devices that provide a digital output for an estimation phase, examples being the analog devices 1574 and 2581, as described in the respective data sheets available from Analog Devices, Norwood, Massachusetts, USA. However, these devices are unsuitable for strip interpolation and they have poor frequency response at high tracking rates.

It will be appreciated that although the above preferred embodiment is primarily intended for an interferometric device, it could also be used where a phase-to-digital converter is required that operates down to low frequencies or DC voltage.

Preferably, the control signal is a digital signal, whereby the control signal can provide a direct digital output measure of the phase. Preferably, the means for providing the control signal is a simple digital counter which is incremented when the phase difference between the reference signal and the signal derived from the interferometer becomes greater than a predetermined value; this provides a fast-acting digital phase output circuit with a simple structure.

The control signal is conventionally supplied to a digital function generator, which may simply be a ROM, which generates a corresponding digital reference signal value, which is then converted to an analog signal by a digital-to-analog converter. In such arrangements, relatively accurate and fast-responding digital-to-analog converters are substituted for conventional analog-to-digital converters.

A particular problem can arise when a sudden vibration or mechanical shock is applied to the probe. At very high probe speeds, the interpolator may not be able to follow the signal from the interferometer. However, a fringe counter circuit can generally operate at considerably higher speeds and thus approximately track even when there are rapid probe movements caused by random vibration.

It is preferable to ensure that the interpolator is phase synchronized with the strip counter so that each ring count resynchronizes the interpolated phase and the interpolator and counter agree.

In the TALYSURF (TM) device indicated above, the probe is held in contact with the surface over which it passes by gravity. The probe must therefore point directly downwards towards the surface to be measured, which limits the application of the device to measuring surfaces in situ from below or from the side.

In one embodiment of the invention, a surface or profile measuring device is therefore provided with a probe and a device for pressing the probe against the object to be measured.

Known arrangements in which the probes are prestressed are known from the documents US-4,669,300, GB-2 085 156 and GB-1 429 973.

The device could simply be a tension or compression spring arranged to push the probe in one direction against the surface However, the force exerted by such a spring is proportional to the deflection of the probe. Therefore, it is preferred that the device exerts a substantially constant pressing force on the probe. Alternatively, the pressing force is such that vibration of the probe is dampened; e.g., it is related to the probe speed.

The device may also be designed to raise the probe from the surface if desired.

The device may include an electromagnet having a relatively adjustable coil and a core which exerts a substantially constant force which can be controlled by a current applied to the coil. In an alternative construction, the probe may be biased to an initial deflection position, e.g. by providing a pair of pressing devices which exert pressure on the probe in opposite directions.

Other aspects and preferred embodiments of the invention will become apparent from the description and claims as they follow.

Embodiments of the invention will now be described, by way of example only, with reference to the accompanying drawings, in which

Fig. 1 shows schematically a known form of surface measuring device;

Fig. 2 shows schematically a known form of roundness measuring device;

Fig. 3 shows in more detail a known type of measuring device for use in Figs. 1 and 2;

Fig. 4 shows schematically the elements of a grating interferometer,

Fig. 5 shows schematically the arrangement of a preferred embodiment of the invention for use in the device of Fig. 1 or 2;

Fig. 6 shows schematically a part of Fig. 5 in more detail;

Fig. 7 is an exploded view of a portion of Fig. 6;

Figs. 8A and 8B illustrate the effects of a curved diffraction grating and a compensating lens therefor;

Figs. 9A-9C illustrate output signals produced by the device of Figs. 6 and 7;

Fig. 10 shows schematically a diffraction grating for use in the embodiment of Figs. 5 to 7;

Fig. 11 shows a schematic cross section through a part of the grating surface of the diffraction grating shown in Fig. 10;

Fig. 12 shows schematically the manufacture of the grid of Fig. 10;

Fig. 13 shows schematically the grating produced by the process of Fig. 12;

Fig. 14 is a detailed, scaled side view of the prism used in the device of Figs. 5 to 7;

Fig. 15 illustrates schematically a first stage in the alignment of the device of Fig. 6;

Fig. 16 illustrates a second stage in the alignment of the device of Fig. 6;

Fig. 17 shows schematically a part of Figs. 6 and 7 in more detail, according to a preferred embodiment of the invention;

Fig. 18 shows an embodiment alternative to that of Fig. 17 ;

Fig. 19 shows an additional embodiment to that of Figs. 17 and 18;

Fig. 20 shows schematically a first alternative embodiment to that of Figs. 5 to 7;

Fig. 21 shows schematically a second alternative embodiment to that of Figs. 5 to 7;

Fig. 22 shows schematically a third alternative embodiment to that of Figs. 5 to 7;

Fig. 23 shows schematically a fourth alternative embodiment to that of Figs. 5 to 7;

Fig. 24 shows schematically a fifth alternative embodiment to that of Figs. 5 to 7;

Fig. 25 schematically shows a first signal processing circuit for use in the above embodiments;

Fig. 26 shows schematically a second signal processing output circuit for use with that of Fig. 25;

Fig. 27 shows in more detail the construction of a counter which forms part of the circuit of Fig. 26;

Fig. 28 shows schematically in more detail a part of the circuit of Fig. 27;

Fig. 29A-F shows schematically signals at points of the circuit of Fig. 28;

Fig. 30 shows in more detail a part of the circuit of Fig. 29;

Fig. 31 shows schematically the general construction of an interpolator circuit forming part of the circuit of Fig. 26;

Fig. 32 shows part of the circuit of Fig. 31 in more detail;

Fig. 33 shows schematically the structure of an interpolator according to a first embodiment;

Fig. 34 shows schematically an interpolator circuit according to a preferred embodiment of the invention, together with a counter circuit;

Fig. 35A-F shows schematically signals at different points of Fig. 34;

DEVICE FOR SURFACE AND PROFILE MEASUREMENT

Referring to Figure 1, a surface measurement system generally includes a support frame 100 having a base 100a and a post 100b to which a feed unit 110 is mountable. The feed unit can be mounted at various vertical positions on the post 100b. Extending from the feed unit 110 is a support member or arm 120 which carries a downwardly directed pick-up needle or probe 130 having a pin with a tapered tip.

For measuring a linear profile across a surface of an object 140, the feed unit 110 is provided with a precision motor for pulling the support arm 120 and probe 130 linearly inward across the object 140. The feed unit 110 is provided with an output device 150 for transmitting signals in an appropriate format to a display or processing device such as a computer terminal or workstation 160.

The signals typically include a signal representing the height of the probe 130 (and thus that of the surface of the object 140) and the path along the object over which the motor moved the probe.

According to Fig. 2, the feed unit 110 for measuring the eccentricity or the roundness of e.g. shafts does not require a linear drive motor; instead a rotary motor is provided for rotating the object 140, and correspondingly, an output device 170 is provided which represents the rotational position of the object.

KNOWN INTERFEROMETRIC DEVICES

According to Fig. 3, a feed unit 110 in a known interferometric Measuring device a helium-neon laser 111, an axis 112 arranged parallel to the surface to be measured (in other words, substantially horizontally), a carriage unit 130 with a motor which acts on the axis 112 in order to adjust the carriage 113 along the axis 112 (by means of a reduction gear), a carriage position sensor (not shown) for generating a signal representative of the position of the carriage 113 along the axis 112, and a pickup tube 114 for forming a relatively light-tight jacket which is firmly connected to the carriage 113.

Light guides 115a, 115b direct the beam from the laser 111 to the pickup tube 114. The pickup needle 130 is attached to the end of the holding arm 120, which is rotatably attached to the pickup tube 114 by means of a rotation axis 121.

The support arm 120 extends beyond the axis of rotation 121 and a reflector 122 having a corner reflector is mounted at the other end of the arm 120. The corner reflector defines the end of one arm of a Michelson interferometer as shown, which requires no further description, and a beam splitter and quarter wave plate assembly 123 is mounted to produce a pair of output signals that are 90° out of phase. The signals are supplied to respective photoelectric detectors which produce respective sine and cosine electrical signals for fringe counting. The volume of the laser and optical system required in this type of interferometer is shown in Fig. 3.

As the probe 130 is pulled across the surface of the object 140 by the carriage 113 running along the axis 112, the probe 130 is pressed against the surface by gravity so that it rises and falls according to the surface; the reflector 122 falls and rises accordingly, changing the length of a path in the Michelson interferometer, causing fringes to pass across the beam splitter assembly 123 and the signals at the detectors to vary. The detector signals are processed by the display or output device 160 to provide a representation of the profile being measured or corresponding data.

GRATING INTERFEROMETER

Referring to Fig. 4, a brief description is given for the application of the known type of grating interferometer for position measurements.

A diffraction grating 200 comprises, in the prior art, a plate having a number of parallel, straight ridges on a surface spaced apart by a pitch d defining a diffraction grating. The grating 200 is movable normal to the line of ridges in the grating plane. The grating 200 is illuminated by a light source 210. If the grating material is transparent, illumination may be from the side of the plate other than that with the grating. The light from the light source 210 is preferably that from a laser, although since no large coherence length is required, it is possible to use any other collimated source of well-defined frequency. If the line spacing on the grating 200 and the wavelength of the light from the light source 210 are of the same order of magnitude, diffraction occurs; e.g. For example, with a light wavelength of 670 nm (produced by a laser diode) and a grating spacing or step size of 1/1200 mm = 833 nm, a pair of strong beams +1st and -1st are produced at angles +/- Θ to the normal on the grating 200. Θ in this example is approximately 54º.

In the path of the two beams, a pair of gratings 220a, 220b are positioned so that the beams are brought together at a spatial point at which a combiner 230 is positioned. The mirrors 220a, 220b need not be parallel; in an alternative configuration, each may be tilted to its respective beam path by an equal angle, but opposite in direction to the other. Furthermore, the mirrors 220a, 220b need not be equidistant from the grating 200. It is preferred that the paths traveled by each beam through the mirrors to the combiner 230 be of equal length so that a light source with a short coherence length can be used.

The compositor 230 is conveniently a dichroic beam splitter prism, but could alternatively be a half-mirrored mirror. At least a portion of one of the beams is reflected at a reflection surface 532 within the compositor 230 so as to coincide with the path of the second beam transmitted through the layer 235. The output beam C from the compositor thus contains the two first order diffraction beams from the grating 200.

However, when a dichroic combiner 230 is used, the transmitted and reflected portions of the combined beam C have different polarizations. Therefore, an analyzer 240a with a polarizing film is provided in the path of the combined beam C with the polarization axis lying between the polarizations of the reflected and combined portions of the beam to allow the same portion of each to pass through; after passing through the analyzer 240a, the combined beam therefore contains a beam whose amplitude is the sum of the amplitudes of the two diffracted beams from the grating 200. A photoelectric detector or some other optical pickup or detector is therefore positioned in the beam behind the analyzer 240a.

If the compositing device 230 is a dichroic prism, a second composite beam D can also be produced, similarly comprising both diffraction orders but with respective right-angled polarizations, and a second analyzer 240b having a polarizer with a polarization intermediate that of the two beams is provided accordingly to produce a second output signal corresponding to the sum of the amplitudes of the two diffracted beams. This second output beam can be used to provide a second signal lagging the output of the analyzer 240a by a fixed phase lag, e.g. 90°, if a transmission plate 250 is provided with a thickness corresponding to an optical path length of a corresponding fraction of the wavelength of the light source 210, e.g. a quarter-wave plate.

At a predetermined transverse position, the two first order beams produced by the diffraction grating 200 are in phase, and therefore the beam at the output of the combiner 240a is a sine wave having twice the amplitude of one of the diffracted beams. If the finite length of the grating 200 is neglected, this grating 200, when translated transversely by a distance corresponding to one grating period or an integer number thereof, appears optically identical; therefore, the amplitudes of the combined beam behind the analyzer 240a are equal and maximum when the grating is at positions spaced apart by one grating period.

If, however, the grid 200 is shifted transversely by a distance that does not correspond to an integer number of the grating periods, the phases of the two diffracted beams are shifted by equal but opposite amounts. In other words, the phase of the beam of one order is shifted forward and that of the other is shifted backward. The effect of the phase shift is that the two components of the composite beam C are out of phase by twice the phase shift, which reduces the amplitude of the composite beam output from analyzer 240a. If one beam is shifted forward by 90º and the other is shifted backward by 90º, the two components of the composite beam are 180º out of phase with each other and the amplitude of the composite beam behind analyzer 280 is close to zero (or in any case the minimum). If the grating is shifted by half its step size, the phase of each diffracted beam is shifted by 180º, so that the phase difference between the two in the composite output of analyzer 240a is 360º - in other words, they are back in phase.

In summary, a detector positioned to measure the amplitude of the beam behind the analyzer 240a as the diffraction grating 200 is transversely translated would measure amplitude maximums at grating positions spaced apart by half the grating period, separated by intervening dark stripes or amplitude minima; the beam amplitude is generally a sinusoidal, positive-going function of the transverse position of the grating 200.

Thus, it can be seen that the grating interferometer illustrated above measures the phase change in a light beam diffracted by a grating that is transversely translated; therefore, it can be seen that instead of the first order diffracted beams, a pair of higher order diffracted beams could be used. It would also be possible to derive the phase shift of the diffracted beams in a manner other than using a pair of diffracted beams of the same but opposite order; e.g., the incident beam from the light source 210 could be used as the reference phase source.

However, the advantage of using a symmetrical arrangement of the type illustrated in Fig. 4 is that the optical path lengths travelled by the two beams can be made practically equal. In order to produce interference, the optical path length between the beams travelled by the two beams are kept below the coherence length of the light source. Accordingly, while it would be possible to use different optical path lengths if a laser source with a high coherence length (such as a helium-neon laser) were provided, the use of symmetrical, approximately equal optical path lengths enables the use of short coherence light sources such as semiconductor laser diodes which may have a coherence length of only a fraction of a millimetre. If the two path lengths are sufficiently similar, a substantially incoherent light source could be used instead of a laser.

For example, a red LED (light emitting diode) operating at a wavelength of, say, 650 nm could be used, the beam from which would be filtered by a narrow band optical filter with a bandwidth of about 1 mm; this would provide a coherence length of about 400 µm, which is sufficient when using the above embodiment, since the path lengths are substantially the same, at least to this accuracy. In order to collimate the beam from the LED, an aperture is also present in this embodiment. It is preferred to use a monochromatic source such as an LED rather than a completely incoherent or broadband source, since this avoids the appearance of colored fringes in the interference pattern.

If a pair of mutually phase-shifted output beams C, D are generated as shown in Fig. 4, the amplitudes of the two beams can be processed to derive an indication of the direction of translation of the grating 200.

The profile of the ribs on the grating 200 affects the relative strengths of the diffracted beams of different orders; an approximately sinusoidal profile is preferred when, as here, strong first order diffracted beams are required. Rather than forming the diffraction grating in the form of a plurality of ribs, it is possible to instead provide an amplitude grating having alternating reflective and absorbing stripes.

The effect of varying the wavelength of the light source 210 is to change the diffraction angle Θ of the diffracted rays according to sin Θ = m λ/d, where λ is the wavelength. In case a change in wavelength occurs, the mirrors 220a, 220b, the compositing device 230 and the analyzers 240 etc. must be made sufficiently wide to accommodate the extended diffraction angle range.

Finally, it should be noted that the effect of moving the grating 200 perpendicular to its plane is to shift the diffracted beams together in corresponding directions; this would therefore eventually shift the beam paths so that they would no longer include one or more of the optical components of the interferometer. However, the effect of angularly rotating the grating 200 is much more pronounced, since the angles of the diffracted beams are shifted in opposite directions. Therefore, the beams quickly become out of coincidence at the compositor 230, resulting in failure of the device to operate. Thus, the operation of the interferometer shown in Figure 4 is considerably more sensitive to angular misalignment between the grating 200 and the other components of the interferometer than to either longitudinal displacement of the grating 200 or wavelength displacement of the light source 210.

FIRST EXAMPLE OF IMPLEMENTATION

Referring to Fig. 5, which corresponds substantially to Fig. 3 and in which like parts are labeled accordingly, a feed unit 110 includes a transverse axis 112 and a carriage 113 which is adjustable along the axis 112 by a motor (e.g., a DC motor driven by a reduction gear). A probe or pickup needle 130 is provided near one end of a support member or arm 120 which is attached by a pivot bearing 121 to a pickup body assembly 114 which is fixedly connected to the carriage 113. Also provided is a signal processing circuit 155 which receives electrical signals from the body 114, as discussed below, and an output port 150 which provides output signals from the signal processing circuit 155. Also included, omitted for clarity, are a power supply and control lines for the carriage motor and a position output circuit that provides the carriage position on the axis 112, along with power supply lines for the body 114 and the signal processing circuit 155.

According to Fig. 6, within the body 114 there is a light source 310 with a laser diode having a wavelength of approximately 670 nm and a collimator lens present in the beam. The support arm 120 extends over the pivot bearing 121 into a portion 123 at the end of which is mounted an optical component having a curved surface whose curvature corresponds to that of an arc of a circle whose center lies in the pivot bearing 121. On the curved surface there is a diffraction grating having a plurality of parallel diffractive features inclined parallel to the axis of rotation 121. Light from the light source 310 is directed straight through a prism 317 normal to the surface on the diffraction grating 300. Two first order diffraction beams produced by the diffraction grating 300 enter the prism 317 discussed in more detail below which produces two output beams each of which passes through a respective output analyzer 340a, 340b having a beam splitter prism. Beam splitter prism 340b is preceded by a quarter wave plate 350. On two surfaces of each analyzer beam splitter 340a, 340b are respective detectors 341a, 342a, 341b, 342b (not shown in Fig. 6). Each detector includes a photodiode responsive to the amplitude of the incident light to produce a corresponding electrical output signal.

Also present in the path between the light source 310 and the diffraction grating 300 is a lens 318 which acts to collimate the collimated beam from the light source 310 to reduce the divergence created by the curvature of the diffraction grating 300, as discussed in more detail below.

Further, connected to the support arm 123, there is a biasing force assembly 400 comprising a linear solenoid coil 410 surrounding a linear solenoid armature or pole piece 420 connected to the support arm 123 so as to exert a tensile or compressive force thereon in accordance with the current supplied to the coil 410.

The pitch of the grating and its distance from the axis of rotation 121 are to some extent related, since the ratio of the movement of the measuring instrument to the movement of the probe is determined by the ratio of the arm sections 120 and 123, or in other words, the radial deflections of the probe 130 and the grating surface 301. The pitch of the grating is determined in some way by the wavelength of the available light source and by limitations of the process used to make the grating.

For a given grid size, the distance traveled by the grid (and hence the number of fringes produced by this displacement) is relative to that traveled by the probe in proportion to their respective distances from the axis of rotation 121. For this reason, a relatively long arm section 123 is desirable. On the other hand, if the arm 123 is too long, the inertial force about the axis of rotation 121 will dampen the response time of the measuring device. In the illustrated embodiment, the section 123 (from the axis of rotation to the grid surface 301) is chosen to be approximately half the length of the support arm 120. For a surface measurement, the length of the support arm 120 is typically 60 mm.

Referring now to Figure 7, the operation of the device shown in Figure 6 will be discussed in more detail. The laser diode 311 is activated to produce an output laser beam which is typically collimated by a collimator lens 312. Typically, the beam produced by the laser diode and collimator lens has a width of approximately 2 mm. The collimated beam passes through a transparent half-wave plate 319 which is present to allow adjustment of the polarization direction of the beam. The light beam is passed through a cylindrical lens 318 which collimates the collimated beam. As shown in Figure 8a, in the absence of the cylindrical lens 318, the collimated beam would produce diverging diffracted output beams when diffracted by the convex curved diffraction grating 300.

By attaching the cylindrical lens 318, an appropriate convergence is created in the input beam so that the diffracted rays from the diffraction grating are collimated, as shown in Fig. 8b. The lens 318 can also correct any divergence or convergence in the beam from the light source 310.

Then the beam is received normally at the end face of the prism 317 and it travels along the central axis of symmetry of the prism 317 and impinges perpendicularly on the surface of the diffraction grating 300.

Since the surface 301 carrying the diffraction grating is arranged on a cylindrical surface centered around the axis of rotation 121, the portion of the surface 301 onto which the light beam strikes (or more precisely, the tangent to this surface) is always perpendicular to the light beam, regardless of the orientation of the rotary arm 123 to the axis of rotation 121. In contrast, If the grating were present on a flat surface, the angle between the grating and the light beam would shift upon rotation about the axis of rotation 121, which would also apply to the distance from the light source 310 and the prism 317.

A pair of first order diffracted beams are produced with an angle e that depends on the illumination wavelength λ and the pitch or distance between grating lines; for a pitch of 1200 lines/mm and an illumination wavelength of 670 nm, the angle of diffraction θ against the normal axis on the grating is approximately 54º. The two diffracted beams enter the rear flat surface of the prism 317 and are refracted by an amount that depends on the refractive index of the same. The refracted beams each strike a respective side surface 320a, 320b of the prism and, provided that their angle of entry therein is greater than the critical angle for total internal reflection with respect to the material of which the prism is made, are reflected back to the center of the prism. The inclinations of the surfaces 320a, 320b to the center of the prism are equal and opposite, so that the two rays meet at the same point in the center of the prism.

A dichroic layer 335 is disposed along the longitudinal center plane of the prism, as is conventional, to respond to an incident beam by transmitting a portion of it in a first polarization plane and reflecting a portion of it in a second polarization plane (polarizations S and P).

The planar layer 335 therefore reflects a portion of each diffracted beam in accordance with the transmitted portion of the other to produce composite output beams. However, in each composite beam, the reflected and transmitted portions exhibit different polarizations and their amplitudes are therefore not additive. Each beam exits the prism 317 through an end face inclined normal to the beam path. One beam enters an analyzer 340a; the second enters a quarter wave plate 350 before entering an analyzer 340b.

Each analyzer 340 includes a further beam splitter prism, each of which comprises a cubic prism cut along a diagonal plane, including a dichroic layer structure between the two halves thereof. The effect of the dichroic layer in the 45° diagonal plane of each analyzer is to act as a beam splitter, passing a portion of an incident beam and reflecting a second. The rotational orientation of the diagonal plane of each beam splitter 340a, 340b is selected so that both the reflected and transmitted beams produced by it pass an equal portion of the S and P polarizations into the output beam of prism 317, thereby containing an equal portion of each of the diffraction orders from diffraction grating 300. Beam splitter prisms 340a, 340b are therefore rotationally inclined at 45° to the planes of prism 317 to which they face. Conveniently, the beam splitter 340a is adhesively cemented to one end face of the prism 317, and the quarter-wave plate 350 and the beam splitter 340 are cemented to each other in that order.

A photodetector (e.g., a photodiode) 314a, 314b is provided to receive the reflection beam from each respective analyzer 340a, 340b, and another detector 342a, 342b is provided to receive the transmitted output beam of a respective beam splitter 340a, 340b. The reflected output beam is in each case 180º out of phase due to the reflection.

Referring to Fig. 9, a graph is shown showing the output of each detector for a rotational displacement Θ of the grating 300 equal to half the displacement between adjacent grating lines. Although ideally the output of each detector is sinusoidal with respect to the grating displacement ranging between zero and a maximum value, in practice it is found to vary between a maximum value and a non-zero minimum value (due to ambient light and the finite beam and grating size, among other factors). The minimum and maximum intensities are equal in the reflected and transmitted beams, or can be made equal, but as stated above the sinusoidal component is 180° out of phase.

In this embodiment, rather than using one of the reflected or transmitted output signals directly, the signal processing following the counting of the fringes is simplified by having a pair of detectors 342a, 341a or 342b, 341b and subtracting their electrical output signals from each other to provide the subtracted signals illustrated in Fig. 9C. The result (not to scale) shown) is a sinusoidally varying signal with twice the amplitude of the sinusoidal components of the individual detector output signals, centered around zero because the DC components of the two signals are canceled upon subtraction.

Curved Grid 300

Referring to Figure 10, the grating 300 in the illustrated embodiment comprises a glass block, the bottom of which is approximately rectangular, measuring, for example, 6 mm x 4 mm, and the top of which is ground or cast to a precise cylindrical profile with a radius equal to the height of the block (typically 5 mm) plus the length of the arm 123 to which it is attached. This may be, for example, 30 mm. Referring to Figure 11, the block carries on its curved surface a diffraction grating having a pattern of ridges spaced apart at a pitch of typically 0.8333 µm (1/1200 mm). A sinusoidal profile is preferred to provide strong first order diffracted beams. The ridged surface is coated with a reflective layer, such as aluminum, to provide a reflective grating.

As shown in Figure 12, one way of creating such a grating is to coat the curved surface of the glass starting part with a layer of a photocurable compound and to direct thereon a pair of inclined laser beams having a wavelength of the order of the required grating. This hologram technique produces a well-defined interference pattern with a sinusoidal intensity distribution, and this produces a corresponding exposure fringe pattern in the photosensitive layer. When the exposure process is complete, the surface is etched or washed to remove either the exposed or unexposed areas of the photosensitive layer, leaving the rib pattern. The patterned surface, with the ribs, can then be coated with aluminum by any suitable process. Alternatively, the rib pattern could be used as a mask through which a selective etching process is carried out.

According to Fig. 13, when this technique is used, the pitch of the curved grating is only absolutely correct at the apex of the substrate and increases slightly towards the edges of the component; in the example given If the pitch at the center is 0.8333 µm, the pitch at the edges is 0.8372 µm (approximately 0.5% higher). Since the beam width is on the order of 2 mm, the variations in the grating pitch that lead to an increase in each diffracted beam are a fraction of 1%, but still affect the performance of the grating somewhat.

Increasing the grating pitch toward the edges of the grating also slightly shifts the diffraction angle. However, since the angle is shifted for both diffracted beams, they still have equal path length and coincide within the prism, so this effect is insignificant, provided that the optical components and detectors are all of finite length. Finally, increasing the grating pitch toward its edges makes the relationship between the number of fringes detected by the detectors and the angles through which the grating rotates slightly nonlinear toward the edges. However, the relationship between the vertical motion of the probe and the rotation of the grating is nonlinear in the opposite sense, so this effect is somewhat mitigated.

Any remaining non-linearity can be well measured and characterized for given gauge dimensions or derived by performing a calibration procedure for the gauge. It is simple to provide correction circuitry to correct the output signals derived from the interferometer, as explained in more detail below, or correction can be performed by a computer or other device 160 to which the gauge is connected.

Nevertheless, it is believed that the performance of the device is improved if a curved grating 300 is provided in which the step size variation is reduced below this level (and preferably eliminated).

Prism 317

Referring to Fig. 14, the structure of the prism 317 will now be discussed in more detail.

The prism is symmetrical around the dichroic layer 335 and comprises a A pair of inclined sides 320a, 320b and the base and top surfaces 360, 361 normal to the dichroic layer. In depth, into the sheet in Fig. 14, the prism 317 preferably corresponds to the grating 300 or is somewhat wider; that is, it is at least 4 mm deep. A pair of exit planes 370a, 370b are provided so that they are arranged at an angle to the central plane 335 so that they are normal to a ray which is totally internally reflected about a point in the central plane 335 at the opposite one of the surfaces 320a or 320b.

Conventionally, the prism is manufactured in the form of two components above and below the center plane 335 in Fig. 14 and assembled at the center plane 335.

The material of the prism is preferably glass, for example with a refractive index of 1.51, for dimensional stability and ease of manufacture. In this case, the dimensions of the prism shown in Fig. 14 can be as follows in order to interact with a grating of period d = 0.8333 um so that λ = 660 nm - 680 nm the diffraction angle Θd = mλ/d, with m = +/-1:

The multilayer dichroic coating on level 335 is calculated and deposited in a known manner to provide a surface where the beam incident thereon is split into a reflected and a transmitted component having different polarizations with substantially the same amplitudes.

The sides 320a, 320b are inclined inward at equal angles so that the angle of incidence of the diffracted rays after reflection on the sides 320a, 320b onto the beam splitting center plane 335 is close to 45º.

In this case, the coating within the beam splitting layer 335 is relatively simple.

By carefully selecting the refractive index of the material of the prism 317 and the distance between the prism 317 and the grating surface 301 as well as the angles of the diffracted rays (which in turn are determined by the wavelength of the source of the incident light and the pitch of the diffraction grating 300), it would also be possible to form a prism 317 with parallel sides 320a, 320b and an angle of incidence of 45° in the middle layer 335.

A suitable coating in the middle layer 335, for an angle of incidence of 45°, is selected to transmit a beam of polarization P and reflect a beam of polarization S with approximately equal intensity and to have very low transmission in the plane S and reflection in the plane P. This is achieved by arranging alternately interleaved layers with two different refractive indices; e.g., eight layers of MgO with a thickness of 216 nm interleaved with seven layers of MgF2 with a thickness of 264 nm. The layers are deposited, e.g., by chemical vapor deposition or any other suitable process to provide a relatively homogeneous transparent layer. A coating of the same construction can be used within the beam splitter prisms 340a, 340b.

Mechanical assembly and alignment

In one embodiment, the probe body 114 is in the form of two body parts 114a, 114b. A first part 114a contains the laser light source and the beam splitter prism 317. A second part 114b contains the grating 300 and a rotation mount 121 for the support arm 123. The prism 317 is fixedly mounted within the first body part 114a, and the laser and lens assembly 311, 312 and half-wave plate 319 are mounted therein so as to allow limited movement in all three planes and axial rotation.

The first step is to center the light beam from the laser along the optical center axis of the prism 317. To achieve this, the first body part 114a is mounted in a jig as shown in an alignment telescope 1000 having self-reflection capability and an x/y aligned reticle. The cylindrical lens 318 is initially absent from the assembly. Using the self-reflection adjustment of the telescope 1000, the body 114 is aligned in the chuck so that the base 360 of the prism 317 is normal to the telescope axis. The position of the body 114a within the plane normal to the telescope axis (the x/y plane) is then adjusted until the center of the prism face 360 is aligned with the telescope axis.

Then the telescope is focused on the surface 360 of the prism. The laser beam 311 is irradiated, providing an elliptical beam spot. The laser is rotated until the ellipse lies in the y/z plane.

Then the telescope 1000 is focused to infinity. The position of the laser 312 within the body 114a is then adjusted in the x and y directions until the spot of light from the laser 312 arrives at the center of the crosshairs of the telescope 1000. Then the laser 312 is positionally locked within the body 114a, e.g., by cementing the laser to it using an adhesive. Then the cylindrical lens 318 is inserted into the body 114a within the beam path of the laser beam and its position in the y direction is adjusted until a focal line is observed through the telescope 1000. Then the alignment of the lens 318 is adjusted until the beam lies along the x axis of the crosshairs within the telescope. Now the cylindrical lens 318 is correctly positioned and it is cemented or otherwise secured in place.

Now the first body part 114a is correctly aligned. The next step is to assemble the two parts 114a, 114b. The two parts 114a, 114b are connected together by a connecting mechanism that initially allows some movement in the x and y directions and rotation about the z axis. For example, the two parts may have facing flanges that elastically press against each other. The second body part 114b is mounted in a jig and the position of the first part 114a is adjusted in the "x" direction while observing the spot of the laser beam until the beam lands centrally on the grating 300.

One of the beam splitter analyzers 340a is removed and the two beams as they are reflected by the dichroic layer 335 or these are observed, either visually using a piece of cardboard or using an optical power meter. The half-wave plate 319 is rotated until the intensities of the two beams are equal.

Then, the body portion 114b is translated in the "y" direction until the two beams are observed to be aligned in the y direction. At this point, the laser beam is incident normally on the surface of the grating 300.

Then the body portion 114b is rotated about the "z" axis to align the grating lines with the beam splitter prism. The body portion is rotated until the two beams imaged on the piece of cardboard are seen to overlap. The two beams should now interfere and bright and dark stripes should be observed when the pickup needle or probe 130 is deflected. If no clear stripes are observed, the above steps of aligning in the y direction and rotating about the z axis are repeated.

When clear fringes are seen, the analyzer beam splitter prism 340a is re-cemented in place and the outputs of the pair of detectors 342a, 342b are fed to an oscilloscope. Then the probe 130 is deflected and the fringes produced are observed on the oscilloscope; if necessary, the half-wave plate 319 is adjusted to improve the fringe amplitude and then it is cemented in place. The phases of the two detector outputs are observed and the quarter-wave plate 350 is rotated until a correct phase difference of 90º is achieved; then the quarter-wave plate 350 is cemented in place.

Then the two body parts 114a, 114b are firmly connected to each other, typically by means of an adhesive cement, and now the measuring device is completely aligned.

Probe bias unit 400

According to Fig. 17, in this embodiment, the preload or pressing force exerted on the probe 130 is provided by an actuator instead of by gravity as in the prior art. The actuator 400 comprises an electrical actuator having a linear coil 410 surrounding a pole piece or armature 420 with a rod fixedly connected to the support arm 123 within the rotation axis 121.

The coil 410 can be activated to apply a constant preload force to the arm 123 and hence the probe 130 via the pole piece 420, but the activation current can also be controlled to vary the applied force continuously, if required, e.g. in response to a signal derived from a stress measuring device or an accelerometer in response to a load or acceleration of the probe.

It may be desirable to provide means for manually adjusting the force applied, for example to apply a relatively high force to relatively rigid surfaces to be measured (to ensure good contact with them), but to apply a relatively small force when measuring elastically or plastically deformable surfaces in order to avoid any deformation or damage to them.

It is particularly convenient to control the biasing force electrically, however, the coil 410 and pole piece 420 could be replaced by a mechanical spring (for tension or compression) or, for example, a pneumatic actuator.

By supplying current in the opposite direction to that used to bias the probe, the probe can be lifted from the workpiece surface for return after scanning without damaging it or the surface.

Accordingly, it should be noted that this aspect of the invention is not limited in its application to interferometric surface measuring instruments, but could equally be used, for example, in measuring devices having an inductive pickup probe. Instead of acting on the support arm behind the rotation axis 121, it would be possible to arrange the biasing device 400 in front of the rotation axis 121 to act on the support arm 120 there; however, this reduces the effective length of the arm 120, which limits the usefulness of the device in measuring enclosed areas such as pipes or openings.

Fig. 18 shows an alternative arrangement in which a pair of opposing actuators 400a, 400b, which in this embodiment include coils 410a, 410b activated in opposite directions and corresponding pole pieces 420a, 420b, are connected to opposite sides of the support arm 123. When the activation of the coils 410a, 410b is controlled so that the forces exerted by the two actuators 400a, 400b are in balance (or when, preferably, the forces exerted up and down by the actuators 400a, 400b and gravity are in balance), the probe 130 is loaded to assume a central rest or zero position. This embodiment is applicable to devices for measuring surfaces that can be positioned either above or below the device or at right angles to it.

Referring to Figure 19, in another embodiment, the actuator is designed to provide active damping of the movement of the pickup needle arm; among other advantages, this reduces the severity of vibrations due to external influences such as accidental impacts on the device. The coil 410 is therefore supplied with a current containing a component proportional to the rate of change of the probe deflection and of such polarity that the force exerted thereby by the coil 410 opposes the cursor deflection.

For example, the current may be provided by a first current source 430 providing a constant current Im and a second current source 440 providing such a current that is proportional to the rate of change of the probe deflection signal, as discussed above.

For example, the digital output signal representing the probe deflection Z may be derived from the output signal of a processing circuit, as discussed in more detail below, and which is converted into a corresponding analog signal by a digital-to-analog converter 450 and then differentiated by an analog differentiator 460, which may include, for example, an operational amplifier with a capacitor at its (inverting) input and a resistor in the feedback path to the inverting input.

The signal from the differentiator 460 is then converted into a current signal by a current follower with a current source 440, and its output signal is added to the constant current from source 430 and supplied to coil 410.

It should be noted, however, that a digital differentiator could be used instead of the analog differentiator, in which case the digital-to-analog converter 450 would be subsequently mounted. Other ways of providing a signal to control the actuator 410 to dampen the movement of the pickup needle are equally apparent.

Preferably, a control is provided for varying the force exerted by the actuators; preferably, in this embodiment, not only the magnitude of the force exerted by the actuators but also the balance between the forces exerted by each of them can be varied so that the device can be pushed either downwards to measure a surface below the probe 130 or upwards to measure an area above the probe 130. Again, this embodiment is not limited in its application to interferometric measuring devices.

Alternative optical arrangements

From the foregoing description of an embodiment, it will be clear that a number of variations and substitutions can be made to the optical system with more or less equal effect. Some exemplary such modifications and alternative constructions will now be illustrated with reference to the following Figures 20 to 24, in which parts corresponding to parts described above are numbered accordingly. In Figures 20, 22, 23 and 24, the prism has parallel sides for illustrative purposes; this requires a more complicated coating 335 if the angle of incidence thereon is not 45°, but the parallel sides are easier to manufacture.

Referring to Fig. 20, the optical arrangement of Figs. 6 and 7 can be varied by replacing the reflective diffraction grating 300 described with reference to Fig. 11 with a transparent or semi-transparent grating. The light source 310 generates a beam which is directed through the body 300 of the grating component onto the inner surface of the grating 301, which acts as a transmission grating and directs first order diffracted rays along the same paths as described above with reference to Fig. 7. described into the beam splitter prism 317. The structure of the beam splitter prism 317 is accordingly slightly simplified compared to those shown in Figs. 21 and 23 in that no recess or beam shifting prism 316 is required.

A mirror 500 is provided to direct the beam from the light source 310 into the diffraction grating. Since the curved diffraction grating component 300 now acts as a lens attempting to converge the beam striking it, the collimating lens 318 of Figure 7 is replaced by an expanding lens 518 to compensate for this and produce collimated diffracted beams. Alternatively, since many laser diodes produce diverging beams, the laser diode 311 can be selected so that its divergence compensates for the convergence of the diffraction grating component 300, or the back of the grating 300 can itself be curved for this purpose.

According to Fig. 21, in a particularly preferred embodiment, the arrangement corresponds essentially to that of Figs. 6, 7 and 14, with the exception of the vertex of the prism, the presence of rollers 316 and the location of the lens 318.

According to Fig. 22, the light source 310 and the prism etc. 317 may be rearranged in the arrangement of Fig. 20; again, a diverging lens 518 or a diverging light source 310 is used to compensate for the convergence due to the grating curvature.

According to Fig. 23, the cylindrical lens 318 in the light path between the light source 310 and the diffraction grating 300 can be replaced by a pair of cylindrical lenses 318a, 318b in the path of the diffracted rays, with essentially the same effect.

Referring to Figure 24, the arrangement for directing the incident light from the light source 310 onto the grating 300 can be simplified by omitting the beam shifting prism 316 and replacing the apex of the pyramid 314 with a flat surface normal to the light beam from the light source 310 that is aligned exactly along the center of the center plane 335 of the prism 317. This simplifies the arrangement but requires precise alignment of the incident beam from the light source 317 to the center plane 335 so that the beam splitter properties of the center plane 335 do not affect the incident beam.

In addition to all of the above arrangements, it is also possible to mount the curved diffraction grating on the outside of the rotation axis 321 (i.e., on the support member 120). However, this arrangement would reduce the effective length of the support arm 120 and thus the applicability of the probe to some types of components (such as pipes) where a long support arm 120 is required. Many other variations are also apparent.

SIGNAL PROCESSING

It would be possible to provide analog output ports 150 carrying the analog signals from each of the detectors 341a, 341b, 3421, 342b, or even to provide optical output ports 150 to which the beams as received at the position where the detectors are located are carried out by means of optical fiber cables. Alternatively, the compact construction, as made possible by the relatively small size of the laser and interferometer in the embodiment described above, enables some of the electrical signal processing to be carried out within the same unit, reducing the possibility of electrical or radio frequency interference and improving the versatility of the device. Since the output signals produced by the device described above are similar in nature to those produced by a conventional Michelson interferometer measuring device, the signal processing circuitry used in this type of device can be used in the same way in the preferred embodiments described above. Similarly, the preferred signal processing devices described below could be used in conventional Michelson type interferometers.

However, as shown in Fig. 25, in a specific structure of this embodiment, the output signals of the detectors 341a, 341b, 342a, 342b are subtracted by a pair of differential amplifier circuits to produce sine and cosine signals, and these two signals are supplied to the output ports 150 of the feed unit 110. Then, a large part of the signal processing unit 155 is conveniently integrated with a computer terminal 160.

According to Fig. 26, in a preferred embodiment of this aspect of the invention, the signal processing circuit 155 comprises a strip counter circuit 600 arranged to maintain a count representing the total number of peaks or valleys of amplitude as detected at the output of an interferometric measuring device such as that previously described, and an interpolator 700 arranged to produce an output signal representing the phase position between such peaks or valleys in the output signal. Low resolution data from counter 600 and high resolution data from interpolator 700 are combined and fed to a linearity correction circuit 800 and a scaling circuit 810 and then output either for data processing or to a storage device or to a display 820 as shown. Conveniently, correction circuit 800 and scaling circuit 810 are both provided by computer 160 operating in accordance with a stored control program.

Figure 27 shows the counter circuit 600 in more detail. This counter circuit 600 includes a digital counter 610 with a latched output, such as a 16-bit counter, and a decision circuit 620 that controls the counter 610 to count either up or down. This is necessary because the deflection of the probe, as explained below, is derived from the number of fringes counted, but the fringes counted for movement in one direction must be subtracted from those counted for movement in the opposite direction to derive a deflection reading.

This is especially necessary when, as described above, the strip count is a non-linear function of position.

Therefore, the decision circuit 620 outputs a control signal to the counter 610 indicating whether the latched count value should be incremented or decremented upon the next counted strip. Upon detection of a strip, an enable signal is provided to the counter 610 causing the count value to be incremented or decremented accordingly.

In order for the decision circuit 620 to decide whether the count value should be incremented or decremented, it receives two separate input signals that are spaced apart in phase. The required signal processing is simplified if the two signals are are spaced 90° apart. As can be seen from Fig. 9C, the variation of the signal amplitude depending on the probe position x is substantially sinusoidal, satisfying the relationship y = sin Θ, where Θ is proportional to the path x, and for the grating interferometer discussed above, Θ = 4πx/D, where D is the grating period. The two signals y1 and y2 received by the up/down decision circuit 620 are therefore given by A sin Θ and A cos Θ relative to any starting phase, and are referred to hereinafter as the sin Θ and cos Θ signals, respectively.

Referring to Fig. 28, in a preferred embodiment, the decision circuit 620 comprises a pair of input amplifiers 621a, 621b, e.g., inverting operational amplifier circuits, with a gain K sufficient to scale the input signals to an appropriate level for the following circuitry.

For example, for use with the device described above, the gain may be -2.4, bringing the peak-to-peak outputs of the amplifiers 621a, 621b to 10 V. The amplifiers 621a, 621b may also include bandwidth limiting filters that pass frequencies from 0 Hz up to a maximum limit (e.g., 5 MHz, for reasons discussed below).

The outputs of amplifiers 621a, 621b are provided to comparators 622a, 622b (e.g., inverting comparators set to provide a logic high or logic low output depending on whether the respective input signal is below or above zero, respectively). In general, such comparators exhibit a limited amount of hysteresis between threshold values; if the counter circuit is to cooperate with an interpolator circuit 700 in a manner described in more detail below, it is necessary to constrain the threshold levels to be within the smallest value resolvable by the interpolator circuit 700. For an interpolator resolution of, in degrees, 360/225, the threshold is (10/2) sin(360/256) = 123 mV, where the peak-to-peak voltage is 10 V.

In Fig. 29, Fig. 29A shows a graph of the two input signals for the probe or interferometer position, and this corresponds accordingly to Fig. 9C. Fig. 29B shows correspondingly the changes in the output signals of the Amplifiers 61a, 61b, depending on the probe position x or the phase angle Θ. Accordingly, Fig. 29C shows the output of the comparators 622a, 622b according to the probe travel or phase; the outputs of the comparators show transitions at the zero crossing points of the input signals.

However, Figs. 29A - 29C do not represent the time course of the signals in these circuit stages; each signal can be static or variable depending on whether the probe is stationary or moving. When the probe is moving in a first direction corresponding to an increasing probe path x (e.g. the probe is raised) and thus an increasing phase at a constant speed of the input signals, the output signals of the amplifiers 61a and 61b and the output signals of the comparators 622a, 622b correspond to Figs. 29A - 29C. When the probe is stationary, all signals are constant. When the probe changes direction and travels back (i.e. in the negative x-direction), the corresponding signals at the outputs of the amplifiers and comparators now correspond to a mirror image of those illustrated in Figs. 29A - 29C.

The comparators 622a, 622b are arranged to detect the zero crossing points of the signals from the interferometer; this is an advantageous way of counting fringes as it is essentially immune to variations in the input signal amplitude and provides a well-defined phase point in the signal at which each fringe is counted, increasing the accuracy of the counter. In addition, there are particular advantages when this type of counter is used with an interpolator circuit 700 as the circuit 600 not only counts fringes but also provides well-defined phase reference points.

A first problem is to detect the direction in which the probe is traveling and, accordingly, whether to increment or decrement the count towards a zero crossing. If the zero crossing of the Sin Θ input signal occurs at Θ = 0º and the probe is traveling in the positive x direction (i.e., Θ, the phase, is increasing), the value of the signal immediately after the zero crossing is positive. On the other hand, if the zero crossing is one that occurs at Θ = π, the value of the signal immediately after the zero crossing is negative. Conversely, if the probe is traveling in the opposite direction, the amplitude of the signal immediately after the zero crossing is at �Theta; = 0 is negative, and that immediately after the zero crossing �Theta; = π is positive.

Accordingly, provided it is clear which phase position corresponds to an existing zero crossing, the direction of travel of the probe can be determined by measuring the amplitude of the signal immediately after the zero crossing. To determine whether the zero crossing is a point with Θ = 0 or Θ = π, the value of the cosine signal is examined.

Thus, the circuit of Fig. 28 includes two further elements; a phase resolution circuit 632 for generating a signal indicating whether a present zero crossing corresponds to 0° or π, and a direction detection circuit 624 which indicates whether a present zero crossing is a rising or falling transition, corresponding respectively to a forward or reverse sweep of the probe.

Referring to Figure 10, in a preferred embodiment, the zero-crossing phase detection circuit 623 comprises a pair of latches 625a, 625b with D flip-flops clocked by a 10 MHz clock signal from a crystal oscillator to latch the state of the signal from the comparators 622a, 622b on the positive going edge of the clock signal. The frequency response of the latches is therefore limited to 5 MHz and, to prevent interference from noise, this is the frequency above which the filters within the amplifiers 621a, 621b should cut off according to the selection made.

The outputs of the sin-θ latch 625 are provided to an OR gate 628 via a logic network comprising a pair of AND gates 626a, 626b and a pair of inverters 627a, 627b. The output of the first AND gate 626a follows the normal output of the latch 625a and the output of the second 626b follows the inverted output of the latch 625a.

The inverted output of the cos-θ latch 625b is provided to a third input of the OR gate 628. Therefore, the signal at this input of the OR gate is the inverse of the signal B shown in Fig. 29C. When this signal is low, the difference in the switching times of the AND gates 626a, 626b causes a short period of time in which the outputs of the two are low, and therefore a short duration of a negative going pulse in the otherwise high output of OR gate 628. However, when the output of latch 625b is high, the output of OR gate 628 remains high. The output of latch 625b, which is connected to receive the squared version of the cos θ signal, acts to select those zero crossings corresponding to θ = 0.

The output of OR gate 628, as shown in Fig. 29D, is applied to the CLEAR input of another latch 625, which is again driven high by the clock signal supplied via inverter 630 and is gated by the non-inverted output of latch 625b via an AND gate 631. The output of latch 629 therefore goes low upon the occurrence of a negative-going pulse in the output of OR gate 628 (following a positive-going clock pulse which clocked latches 625a, 625b), and again goes high on the following negative-going edge of the clock signal. The output of latch 629 is therefore a negative going pulse of 0.1 microsecond duration at each zero crossing of the 0° phase of the Sin-θ input signal from the measuring device, as shown in Fig. 29E.

The up or down counting selector circuit 624 must confirm whether the state of signal A is rising or falling when Θ = 0 passes. This can be easily deduced from the output of latch 625a.

The clock signal shown in Fig. 29E and the count up/down control signal shown in Fig. 29F are then provided to the corresponding inputs of a digital counter chip containing the counter 610, e.g., a pair of 74 AS 867 counter chips.

Interpolator 700

Referring to Fig. 31, the interpolator 700 according to this aspect of the invention includes an input 710 receiving the sin Θ signal from the interferometer, an estimation circuit 720 for generating a signal Φ representing an estimate of the phase Θ of the input signal at input 710, and for generating a function of this estimate Φ (here as F(Φ)) and an error generating circuit 730 which generates an output signal δ which is a function of the error between the estimated phase Φ and the actual phase θ of the input signal, this error output signal being fed back to control the estimation circuit 720.

Referring to Fig. 32, a particularly preferred way of providing the estimator and function generator 720 comprises providing a digital counter 721 which stores a count value representing the estimated phase Φ and which is provided with an enable signal (CCK) for changing the count value and a direction indication signal (U/D) indicating whether the count value is to be incremented or decremented. The output of the counter is thus a digital word representing the estimate Φ of the phase Θ of the input signal. The output Φ is also provided to a digital function generator circuit which conveniently includes a read-only memory containing a look-up table which stores, for example, for each value of Φ a corresponding value of the function F(Φ). The data value output from the lookup table 722 is applied to the digital input of a digital-to-analog converter 723, which accordingly produces an analog output signal representing the value of the function F(Φ).

If the circuit shown in Fig. 32, with the function F(Φ) = sin(Φ), were used as shown in Fig. 31, and the deviation signal generator 730 were to operate only as a subtractor, the output signal δ would be the following:

δ; = sin Θ - sin Φ, or

δ; = 2 cos ((�Theta; + Φ) / 2) sin ((�Theta; - Φ) / 2)

Since Θ is approximated by Φ, this leads to the simplification δ = (cos Φ) (Θ - Φ).

In other words, the value of the deviation signal δ is not only a function of the difference between the estimated phase Φ and the actual phase Θ, which is the desired deviation measure, but also of the actual phase Θ or Φ. Accordingly, in order to use this deviation signal as a control signal to increment the counter 721, it would be necessary to provide a test threshold that varies depending on the value of Θ or Φ.

Referring to Fig. 33, an alternative implementation of the principle illustrated in Fig. 31 operates to derive an error signal δ that is a function only of the difference between the input phase Θ and the estimate phase Φ. For this purpose, both the sin Θ and cos Θ input signals from the interferometer are used. As in Fig. 32, a digital counter 721 latches a number representing the value of the estimate phase Φ, and this number comprises the output of the circuit 700. This signal is also fed to the address buses of a pair of ROMs 722a, 722b, each of which stores a look-up table representing a digital number corresponding to the cosine and sine, respectively, of the estimate phase Φ (in other words, a table of constants between -1 and +1). For example, counter 721 may be an 8-bit counter and ROMs 722a, 722b may each contain 256 8-bit numbers, each of which corresponds to each possible value of Φ.

The data buses of the ROMs 722a, 722b are connected to the digital input lines of a pair of respective multiplying digital-to-analog converters 740a, 740b, each of which includes a resistive ladder network with switchability in accordance with the digital input signals to apply a respective resistance to a received analog current and thus attenuate the current by an amount proportional to the digital input signal to the multiplying DAC. The multiplying DAC 740a is connected to the sin Θ input of the interpolator 700 and thus produces an analog output current proportional to sin Θ cos Θ. The multiplying DAC 740b is connected to receive the cos θ input signal for the interpolator 700 and accordingly produce an analog output signal proportional to cos θ sin θ. The deviation signal generating circuit 730 includes a subtractor circuit that produces an output signal proportional to:

sin Θ cos Φ; - cos Θ sin Φ; = sin(Θ - Φ).

This signal is provided to a comparator circuit 725 which produces an output signal for incrementing or decrementing the counter 721 whenever the value of the deviation signal Θ - Φ exceeds a predetermined threshold corresponding to the least significant bit within the phase counter 721, the threshold being given as before by Vpp/2 sin (360(256) = 123 mV for Vpp = 10 V. The counter 721 is preferably designed to count up and accordingly divide the phase of the incoming signal by a number that is sufficiently high for the value of the signal

δ = sin (Φ - Φ)

approximately

Θ - Φ

; for example, if counter 721 is a 4-bit counter, each counter increment corresponds to a phase shift of 22.5º, and comparator 725 causes counter 721 to increment when the estimated phase Φ deviates from the true phase by 22.5º. At 22.5º, sin X is not a satisfactory approximation of X, so counter 721 is incremented at points that do not correspond to precisely regularly spaced phase intervals. On the other hand, if counter 721 is an 8-bit counter, the maximum phase lag of the estimated phase Φ and the signal phase θ occurs at sin θ = 360/256 = 1.4º.

Over the range 0 - 1.40, sin (Θ - Φ) is a good approximation to Θ - Φ.

According to Fig. 34, it may be preferable to store only positive numbers in the look-up tables 722a, 722b. Accordingly, the function values stored in the ROMs may instead be represented by 1 + cos Φ and 1 - sin Φ (with values between 0 and +2). The additional terms sin Θ and cos Θ in the output of the multiplying DACs 740a, 740b are removed by subtracting the input values sin Θ and cos Θ from the corresponding output signals.

An offset of half the least significant bit, 0.7º, can be added to Φ so that the tables store 1 + cos (Φ + 0.7) 1 - sin (Φ + 0.7) to provide rounding up.

Accordingly, the interpolator 700 further comprises a pair of buffer amplifiers 741a, 741b having, for example, inverting operational amplifier circuits with gains of -1. The output signals of the amplifiers 741a, 741b, together with the sin Θ and cos Θ input signals, are summed at the inverting terminal of an inverting operational amplifier 731 having gain of 1, which thereby produces an output signal equal to sin Θ cos Φ - cos Θ sin 4) = sin (Θ - 4)). If the stored values contain an offset, as discussed above, the output signal corresponds to Θ - 4) - offset.

The polarity of this deviation signal (indicating whether the estimate is leading or lagging the desired phase Θ) is sensed by a comparator 751, the logic output of which controls the up/down count input of counter 721.

The deviation signal is also fed to the comparator circuit 725, which comprises a first comparator 725a with a predetermined threshold corresponding to the least significant bit of the counter 721 and a second comparator 725b with an identical threshold but preceded by an inverting operational amplifier 726 with a gain of 1. The comparator 725a therefore produces an output signal when the deviation exceeds a predetermined positive threshold and the comparator 725b produces an output signal when the deviation signal exceeds a corresponding negative threshold.

The output signals of the two comparators 725a, 725b are input to a logic circuit 727 which performs an OR operation to cause a signal EN to change the count value of the counter 721 when either comparator indicates that its threshold has been exceeded. The 10 MHz master clock signal is provided to a synchronization logic circuit 760 which re-times (e.g. latches and delays) the signal from the comparator circuit 725 in such a way that it corresponds to a corrected phase of the master clock. The synchronization logic circuit 760 is connected to the count enable input of the counter 721 to provide for a corresponding increment or decrement of the latched clock count value representing the estimation phase 4).

The synchronization logic circuit 760 is arranged to divide the master clock frequency by 10 before applying a signal from the comparator circuit 725 to the input of the counter 721 so that the counter 721 is incremented only once per microsecond. The reason for this is to prevent switching of the counter 721 due to transient noise levels of the deviation signal, such as those caused by switching operations within the multiplying DACs 740a, 740b; without some means to prevent such transient signals, it would be possible that a change in the count value of the counter 721 would cause toggling transient signals in the output signals of the DACs 740a, 740b, which would then change the count value of the counter again. However, other means to limit the response of the system to such transient signals are possible (e.g. low-pass filtering within the analog circuit path).

In the circuit of Fig. 34 it is seen that the decay times required to allow such transient signals to die out is the factor limiting the speed of the interpolator. A decay time of 1 microsecond, with an 8-bit counter providing 256 values of φ between adjacent stripes, gives a maximum tracking rate of 3.9 kHz (stripes per second).

By providing a digital estimate for phase 4) and using a digital-to-analog conversion to compare it within the analog domain with the input signal, it is possible to provide a cost-effective and relatively inexpensive interpolator compared to devices that use analog-to-digital conversion of the input signal, since cheap analog-to-digital converters are slow and fast analog-to-digital converters are expensive. By providing a digital counter in the digital output device instead of a microprocessor, increased speed is similarly achieved.

Referring again to Figures 34 and 26, it can be seen that counter 600 and interpolator 700 operate essentially independently. Therefore, a problem can arise when the phase Θ of the input signal approaches zero. When interpolator 700 reaches the phase point Θ = 0, its output alternates between 0 and its maximum value (e.g., 256). If this does not occur synchronously with the point at which strip counter 600 detects the zero crossing and therefore counts an additional strip, the combined outputs of interpolator 700 and strip counter 600 will have a deviation of one strip when counter 600 is so operating. Since the circuits are independent, there is any probability that the points at which each registers the zero phase condition are different.

Furthermore, as shown in Fig. 35A, if an offset is stored in the tables and the phase Θ passes through zero, the Θ-Φ offset value may not rise to a level sufficient to trigger the comparators 725a, 725b before the polarity changes. Accordingly, the signal required to increment the counter 721 and cause it to restart at zero is not provided by the comparator circuit 725. To synchronize the counter 600 and the interpolator 700 and to ensure that the interpolator 700 responds to the zero crossing of the phase, the clock signal (CCK) derived within the counter 600 is provided to the synchronization logic 660 as shown in Fig. 32 and inverted as shown in Fig. 33C to provide an additional switching signal provided for incrementing/decrementing the counter 721.

ALTERNATIVE IMPLEMENTATION EXAMPLES

Various modifications to the counter and interpolator circuit described above are possible. For example, instead of a phase counter incremented by the deviation signal Θ, the estimation circuit 720 of Fig. 31 could comprise a free-running counter whose counting frequency is controlled by the deviation signal Θ. However, this has proven undesirable in interferometric measuring devices for surface or profile measurements, since the response of such embodiments is unstable and inaccurate at low frequencies or zero frequency.

Since the deviation signal sin(Θ - 4)) is close to an approximation of the phase deviation Θ - 4), it would be possible to digitize this deviation signal for greater accuracy, adding another, less significant bit to the output of the interpolator. Alternatively, the same accuracy could be maintained by reducing the number of bits in counter 721 to more coarsely divide the phase between strips to a value where sin(Θ - 4)) is still a reasonably good approximation of Θ - 4), and regaining the lost accuracy by digitizing the difference signal. This could increase the maximum tracking rate of the interpolator somewhat.

output

Referring to Fig. 34, where the phase angle counter 721 divides the phase between stripes into a number that is a power of two, the output of the phase counter 721 may be connected to a digital output data bus for the lower order bits 0-7 of a word for which the binary output of the strip counter 610 contains the higher order bits.

When the number of fringes counted is not in exact linear relationship to the path traveled by a probe or the like, as in the interferometric measuring device discussed above, this digital output bus is connected to a non-linearity correction circuit 800 shown in Figure 26, which may simply comprise a read-only memory for the address lines to which the digital output word is applied, thereby producing a corresponding, corrected digital word on its data bus lines.

A scaling circuit 810 with a digital multiplier (e.g., another ROM look-up table) is provided to output the corrected output of the correction circuit 800 in a convenient form in units of distance, if required. In the embodiment described above, the probe distance corresponding to each stripe-to-stripe interval is 0.833 µm, and accordingly each 1/256 phase interval corresponds to 3.25 nm, so the multiplier 810 multiplies by 1/3.25 to convert the corrected number into nanometers.

Although the correction and scaling circuits 800, 810 are shown separately for clarity, in practice they may conveniently contain the same look-up table ROM which performs both correction and scaling. The factor provided by the scaling circuit 810 would be known by calculation from the dimensions and geometry of the measuring device, and the non-linearity correction would conveniently be derived in a calibration phase by measuring the digital output words from the counter 600 and the interpolator 700 against the measurement of known surfaces or profiles.

From the foregoing, it will be appreciated that the digital processing circuit 150 with the fringe counter 600 and the interpolator 700 could be used with other types of interferometric instruments than those previously described, and that the interferometric measuring device previously described could be used with other signal processing circuits.

Furthermore, both the strip circuit and the interpolator circuit could be used separately, although it is particularly advantageous to use a zero-crossing detection strip counter circuit as described above in conjunction with the interpolator circuit described above, as this ensures an accurate phase reference.

Furthermore, the bias force generating means 400 for urging the probe or pickup needle into contact with an object to be measured are applicable to other types of measuring instruments than those described above, e.g., inductive sensing measuring devices. However, when used together, the measuring device described above together with the signal processing output circuit can form an extremely compact unit that can be mounted on a single feed unit 110 without undue stress on its vertical stand, using a low voltage supply that is safer than those previously used for helium-neon lasers, producing a sophisticated digital output signal that provides a dynamic resolution of e.g., up to 1.8 x 106:1.

Claims (45)

1. Apparatus for measuring surface characteristics, comprising a probe (130) for contacting a surface to be measured and moving along the surface, and with a probe holder (120) for supporting the probe (130) so that it can perform a rotary movement about a pivot point (121), characterized by a grating interferometer for providing a measure of the rotary movement of the probe (130), with a diffraction grating (300) arranged on a curved surface which is connected to the probe holder (120) so that it moves with the probe (130) and which has a center of curvature at the point (121) about which the probe can rotate. 2. The device of claim 1, wherein the grid is connected to the probe holder further away from the probe than the pivot point. 3. Device according to claim 1 or 2, with a focusing device for optically correcting the optical effects of the grating curvature. 4. Apparatus according to claim 3, wherein the grating is illuminated by the beam of a beam source and is convex with respect to the beam. 5. Device according to claim 4 or 5, wherein the focusing device comprises a converging lens between the beam source and the grating. 6. Device according to claim 1 or 2, wherein the grating is illuminated by a beam from a beam source so that its beam is focused by it and wherein the grating is concave with respect to the beam and the diffracted rays produced by the grating are due to the divergence of the beam source and the convergence caused by the curved grating are essentially parallel. 7. Apparatus according to any preceding claim, wherein the interferometer comprises means for directing a pair of oppositely diffracted beams of the same order to a connecting means which produces interference between them. 8. Apparatus according to claim 7, wherein the path lengths traveled by the two diffracted beams to the connecting device are substantially equal. 9. Device according to claim 7 or 8, wherein the directing device has two reflecting surfaces. 10. The apparatus of claim 9, wherein the two surfaces comprise outer surfaces of a unitary prism. 11. Device according to one of claims 1 to 6, with a prism into which a pair of diffracted rays of the same order are directed from the grating and from whose surfaces the diffracted rays are reflected towards a connecting device. 12. The apparatus of claim 11, wherein the optical path lengths from the grating to the connector are substantially equal. 13. Device according to one of claims 10 to 12, wherein the reflection from the surfaces is a total internal reflection due to the geometry and the refractive index of the prism. 14. Device according to one of claims 10 to 13, wherein the reflecting surfaces are parallel. 15. Device according to one of claims 10 to 14, wherein the angles of incidence and reflection are due to the geometry and the refractive index of the prism 450. 16. Apparatus according to any one of claims 10 to 15, wherein the connecting means includes a semi-reflective inner surface of the prism. 17. The device of claim 16, wherein the semi-reflective surface comprises a dichroic layer. 18. The device of claim 17, wherein the angle at which the beam impinges thereon is about 45°. 19. Device according to one of the preceding claims with a semiconductor laser diode illumination source. 20. Device according to one of claims 1 to 20 with an illumination source which, when filtered by a narrow-band filter, has a device for emitting substantially monochromatic light. 21. Device according to one of the preceding claims with a biasing device for biasing the probe in a predetermined direction. 22. The apparatus of claim 21, wherein the predetermined direction is toward the surface. 23. The apparatus of claim 22, wherein the predetermined direction points to a specific position. 24. Device according to one of claims 1 to 20, with a biasing device for exerting a force on the probe, so that its vibrations are dampened. 25. The apparatus of claim 24, wherein the force is related to the speed of the probe device. 26. Device according to one of claims 21 to 25, wherein the biasing device includes an electromagnetic actuator. 27. The apparatus of claim 26, wherein the actuator comprises a coil surrounding a core. 28. Device according to one of claims 21 to 27, wherein the biasing means comprises a pair of counteracting biases. 29. Apparatus according to any one of claims 1 to 6, comprising means for combining, with interference, a pair of diffracted beams of different polarizations, said means comprising a beam splitter for forming a reflected and a transmitted output and means for producing a signal responsive to the difference between the outputs. 30. Apparatus according to claim 29, wherein the difference means includes a pair of opto-electrical transducers for generating corresponding electrical signals and means for receiving the electrical signals and generating an electrical output signal proportional to their difference. 31. Apparatus according to any preceding claim, comprising an interpolator for processing signals derived from the output of the interferometer to provide an output signal relating to the angular position between interference patterns of the interferometer output, the interpolator comprising means for generating an estimate signal relating to the angular position, means for Generation of the output signal as a function of the estimation signal, means for generating from the estimation signal a function dependent thereon, and means for controlling the estimation signal generated as a function of the generated function signal and the input signal in order to reduce their difference. 32. The apparatus of claim 31, wherein the estimation signal is a digital signal and the function generating means includes a digital-to-analog converter means. 33. The apparatus of claim 31, wherein the estimate signal represents a digital signal and the control means comprises means for changing the value of the digital signal by a count value when the difference signal exceeds a predetermined threshold to reduce the difference signal. 34. The device of claim 33, wherein the predetermined threshold is constant and the difference signal depends at least approximately linearly on the angular position. 35. The apparatus of claim 34, wherein the difference signal approximates the sine of the difference between the estimate signal and the input signal. 36. The apparatus of claim 31, wherein the control means is arranged not to change the estimate signal in the absence of a change in the input signal to extend the frequency range of the apparatus to DC operation. 37. Apparatus according to claim 31, arranged to receive a pair of input signals of different phase. 38. The apparatus of claim 37, wherein the signal pair includes sine and cosine signals with 90° different phases. 39. Apparatus according to claim 38, comprising means for generating estimated sine and cosine signals, means for generating product signals corresponding to the products of the estimated sine and cosine signals and correspondingly the input cosine and sine signals, and means for generating from the difference between them a signal indicative of the difference in angular position between the input signals and the estimate signals, to control the estimate signal generating means. 40. Device according to claim 39, wherein the estimation signal generating device comprises a digital signal generating device and the multiplication device comprises multiplying digital/analog converters 41. Device according to one of claims 31 to 40, with a device for counting interference patterns or maximums which are detected in the output signal or the output signals. 42. Apparatus according to claim 41 comprising means for synchronizing the estimation signal generator from the counter. 43. Device according to one of the preceding claims with a device for receiving a zero-average output signal of the interferometer and a device for detecting its zero crossings. 44. Apparatus according to claim 43, comprising means for receiving a second input signal having a phase difference to the first, means for determining the phase of the zero crossing from the first and second signals, Means for determining the direction of passage and means for increasing or decreasing the pattern count depending on the direction and phase. 45. A method for measuring surface features, comprising: Moving a probe (130) held by a probe holder (120) along a surface to be measured while allowing a rotational movement of the probe about a pivot point (121), characterized in that a rotational movement of the probe (120) is measured using a grating interferometer with a diffraction grating (300) arranged on a curved surface which is connected to the probe holder (121) in such a way that it moves with the holder (121) and which has a center of curvature located at the point (121) about which the probe (130) is rotatable.